Im trying to train the ssd_inception_v2_coco_2018_01_28 in https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf1_detection_zoo.md with tensorflow 2 but i get the error ValueError: ssd_inception_v2 is not supported. See model_builder.py for features extractors compatible with different versions of Tensorflow. I tried creating a new enviroment with tensorflow 1.15 and installed the object detection api on it using object_detection/packages/tf1/setup.py thinking this would resolve the problem because im installing the object detection api for tensorflow 1 (or so i believe) but when trying to train i get the same error. Im using the files model_main.py, model_main_tf2.py, and legacy train.py but to no avail. I would think that im using the wrong files to train a tensorflow 1 model but i cant find any documentation on that. Any help would be appreciated!
0
votes
Have you tried the script object_detection/builders/model_builder_tf1_test.py ? My guess is you haven't done the installation properly for Tensorflow 1 and that the python files generated with the proto command are not generated well or something like that. If you want to use TF2 check the model zoo corresponding : github.com/tensorflow/models/blob/master/research/…. You could also retry the installation with a docker image for exemple : gilberttanner.com/blog/…. Best of luck!
– dnl_anoj
1 Answers
0
votes
Please go to tf1 installation and follow Python Package Installation. once you run the test below sucessfully, you are good to go for using tensorflow object detection api on you system.
# Test the installation.
python object_detection/builders/model_builder_tf1_test.py
Use tf1_detection_zoo models, If you still come across any error, you can share it.
